As information technology has rapidly progressed in the past ten years, the role of computer network centers such as server farms and server clusters have became increasingly important to our society. The server farms provide efficient data storage and data distribution capability that supports a worldwide information infrastructure, which has come to dominate how we live and how we conduct our day to day business.
A server farm is a group or cluster of computers acting as servers and housed together in a single location. For example, a Web server farm may be either a Web site that has more than one server, or an Internet service provider that provides Web hosting services using multiple servers. In a business network, a server farm or cluster might perform such services as providing centralized access control, file access, printer sharing, and backup for workstation users.
To take advantage of economies of scale, the number of computers hosted in server farms has continued to grow over the past ten years. This has lead to an increasing need for space in which to house the network host units and a consolidation of spaces where they are located. Sites known as co-location sites where numerous networked computers find a home have emerged to meet this market demand. Space for the computers is typically rented at such sites. Rent calculations may be based on the overall space occupied, power consumption and bandwidth handled by the computers occupying the space. Because of the relationship between such factors, it will often be in favor of both a co-location site and computer service provider to maximize both the density and performance efficiency of the computers at a given site. By increasing the density at which computers may be packed into a given area, the service provider benefits as less space is required for a given number of computers; the co-location site benefits since the ultimate bandwidth available in association with the space available may be greatly increased.
Other less apparent benefits stem from conserving the space a host computer occupies. In many instances, it will be economically feasible to forestall the retirement of otherwise outdated host computers since the cost of the space they occupy is relatively lower, thereby justifying their continued service for a period of time. On the other hand, where it is preferred to only maintain the highest-end computers in service, the savings available by minimizing the size of such computers without hindering performance is quite clear. There exists a need for computer systems adapted for realizing these many advantages.
Typically, at a site where numerous computers are connected to a network, the computers are stacked in racks and arranged in repeating rows or cells. Access to the computers is necessary for servicing, upgrading hardware, loading software, attaching cables, switching power on and off, and so forth. The elimination of as much access space as is feasible can increase the density of computer systems that may be provided for a given square footage of area at a site. Consequently, there exists a need to eliminate extraneous access space while still maintaining the use of relatively inexpensive, standard (or more-or-less standard size) racks.
In the market today, a standard rack that is widely used measures roughly 19 inches wide, 30 inches deep and 74 inches high. In at least one co-location site, these racks are lined up in rows of roughly 10–30 units with access doors on each side of a rack. Access aisles are provided on both sides of the rows. Many of the racks are filled with cumbersome computers mounted on sliders which are attached through mounting holes provided in the front and back of the rack. Regardless of the chassis design of the computers (or lack thereof where computers are merely built on open trays with their components uncovered) and how they are mounted to the racks, data devices included in the computer are accessed from the front. Main board I/O's, other I/O's, power cords and such items are typically accessed from the back. It is this latter design aspect which not only results in inefficiency in the amount of access space required, but also in the frequent inefficiencies associated with having to administer services to both sides of a computer. Consequently, there exists a need for computers useable in a network setting that are accessible and fully serviceable from a single side.
As the number of computers in a server farm is increased, two competing factors come into play: consumption of floor space and heat/ventilation management. To increase the number of computers at a given server farm without increasing the density of the computers means one would need more space. As the cost of real estate continues to rise, especially in the urban areas where population density is high, there is a strong incentive to maximize the utilization of a given space. Furthermore, in some existing server farm facilities, there is no more space available for scaleable growth. In such a situation, in order to expand, one would have to absorb the cost of starting a new server farm.
Alternatively, one may try to increase the number of computers that are housed in a given space. In order to significantly increase the density of computers in a given space, one common solution has been to shrink the size of each individual computer in the rack. Another option is to decrease the space between the racks that are holding the stacks of computers.
However, as one increases the density of computers, problems associated with heat dissipation rises exponentially. One of the major causes of electronic component failure is overheating. High performance electronics such as CPUs generate substantial amounts of heat. Additionally, next generation processors and power supplies are emitting substantially more heat as computing requirements increases. Thereby placing further demands on effective heat dissipation. In order for computers to continue to operate properly, appropriate heat dissipation pathways must be provided. Because each computer contains thousands of heat producing electronic parts, as one increases the density of the computers, one must also address the difficult issues of providing proper cooling mechanisms to remove heat from the individual computer nodes and the clusters as a whole.